A Multi-Feature Fusion Model Based on Denoising Convolutional Neural Network and Attention Mechanism for Image Classification

1. Introduction

With the rapid development of computer science and Internet technology, a large amount of data is produced all over the world all the time, such as, text, audio, pictures, videos and so on. Big data contains a lot of information, but this information is difficult to identify and organize manually (Li et al. 2023; Huang et al. 2021). Therefore, how to classify and identify information has been concerned and studied by many researchers. Among all kinds of information, image information is a very important information carrier. In the last century, researchers proposed many excellent image algorithms for texture information in images, which laid a solid foundation for computer vision and pattern recognition. In the 21st century, Convolutional Neural Network (CNN) (Guo et al. 2022) has made great breakthroughs in the field of image classification, so computer vision has entered a new era.

CNNs are a kind of Feed forward Neural Networks with deep structure and convolutional computation. It can better obtain the position space and shape information of image, which is beneficial to image classification. (Ghimire et al. 2022) discovered that Graphics Processing Unit (GPU) played an important role in machine learning (ML), and proposed an efficient CNN training method based on GPU, which greatly improved the computing ability of CNN. (Alex Krizhevsky et al. 2017) proposed AlexNetc network, which adopted ReLU activation function and GPU packet convolution for parallel training. (Teng et al. 2022) applied batch normalization to neural networks, which ensured that the output distribution of each layer in the network was basically stable. At the same time, the network greatly reduces the dependence on the initial parameters and improves the network performance. The following is the classification of feature fusion.